J. Phys. Chem. B 1997, 101, 2591-2597
2591
Light-Induced Charge Separation across Ru(II)-Modified Nanocrystalline TiO2 Interfaces
with Phenothiazine Donors
Roberto Argazzi and Carlo A. Bignozzi*
Dipartimento di Chimica dell’UniVersita, Centro di Studio su FotoreattiVita e Catalisi CNR,
44100 Ferrara, Italy
Todd A. Heimer, Felix N. Castellano, and Gerald J. Meyer*
Department of Chemistry, Johns Hopkins UniVersity, Baltimore, Maryland 21218
ReceiVed: July 1, 1996; In Final Form: October 29, 1996X
Long-lived interfacial charge-separated pairs, [TiO2(e-), D+], have been created by visible light excitation of
Ru(II) polypyridyl compounds anchored to TiO2 particles in the presence of phenothiazine donors, D. The
kinetic aspects of the formation and recombination have been kinetically resolved for analogous colloidal
TiO2 solutions and films. Charge-separated pair lifetimes are shortened in colloidal films due to their high
local concentrations. Less than 1% incident photon-to-current efficiency is observed when PTZ (phenothiazine)
derivatives are employed as donors in regenerative solar cells. Light excitation of Ru(4,4′-(CO2H)2-2,2′bipyridine)2(4-CH3,4′-CH2-PTZ-2,2′-bipyridine)]2+, anchored to TiO2, results in rapid intramolecular electron
transfer from PTZ to the ruthenium metal center which efficiently translates the hole away from the
chromophoric unit to the pendant PTZ group. The net result is the formation of a remarkably long-lived
charge-separated pair, TiO2(e-)|-RuII-PTZ+, that lives for ∼ 300 µs and directly results in an increased
open circuit photovoltage when compared to a model compound.
Introduction
There has been considerable interest in light-induced electron
transfer reactions in nanometer-sized semiconductor clusters.1
In these materials, each colloidal particle acts as an individual
photoelectrochemical cell, sustaining both oxidative and reductive reactions simultaneously at its surface. In more elaborate
assemblies, light-absorbing pigments, catalysts, or redox active
compounds have been anchored to the colloidal surfaces. The
prime motivation is to prepare integrated photocatalytic assemblies capable of converting light into useful fuels or chemical
products. More fundamentally, photophysical studies have led
to keen insights into the factors which govern electron transfer
dynamics at these fascinating interfaces.2
Recently, there has been a growing effort to extend these
studies to the direct conversion of light into electricity.1a,3 In
order to take advantage of nanometer-sized semiconductor
clusters, one must provide an electron pathway for conduction
between the particles. This has been achieved by briefly
sintering colloidal solutions deposited on conductive substrates.
The resultant material is generally a porous nanostructured film
which retains many of the characteristics of colloidal solutions,
but is in a more manageable form. Furthermore, the Fermi level
within each semiconductor particle can be controlled potentiostatically. These materials, and closely related materials
fabricated by electrochemical techniques, display interesting
photoelectrochemical properties that cannot be rationalized by
traditional Schottky junction models.4-7
Our interests were motivated by the impressive solar energy
conversion efficiencies reported by Grätzel and co-workers for
ruthenium polypyridyl sensitizers anchored to porous colloidal
TiO2 films.8,9 Like natural photosynthesis, these materials
convert light into useful energy by efficiently separating charge.
Shown in Scheme 1 are the electron transfer processes that can
be initiated when a pulse of visible light excites a Ru(II)
X
Abstract published in AdVance ACS Abstracts, March 1, 1997.
S1089-5647(96)01939-6 CCC: $14.00
polypyridyl compound anchored to a TiO2 particle. Light
excitation (1) forms metal-to-ligand charge transfer (MLCT)
excited states that are known to rapidly inject electrons into
TiO2 (2) with a quantum yield near unity under a wide variety
of conditions.1 This produces an interfacial charge-separated
pair with the electron in TiO2 and the hole localized on the
ruthenium metal center. There exist at least two possible fates
for this charge-separated pair: recombination to form ground
state products (3) or an electron donor can reduce the oxidized
form of the sensitizer (4). This latter process generally enhances
the lifetime of the electron in the solid and moves the hole from
the surface bound sensitizer to a mobile donor that can do work
external to the semiconductor surface.
In this paper, phenothiazine donors (shown below) have been
employed with Ru(II) polypyridyl sensitizers anchored to TiO2
colloids. Phenothiazines were chosen due to their high solubility
in aqueous and nonaqueous solvents, their well-known outer
sphere one-electron oxidation potentials, and the ability to tune
oxidation potentials over a wide range through substituent
changes.10 Further, phenothiazines become colored when
oxidized which allows the redox chemistry to be followed
spectroscopically. A novel sensitizer with a covalently bound
phenothiazine donor allows (4) to be controlled intramolecularly.
To help bridge the gap between charge separation studies in
fluid solution and the emerging technology of porous nanostructured films, both were explored. We note that preliminary
reports of these studies have recently been presented.11
© 1997 American Chemical Society
2592 J. Phys. Chem. B, Vol. 101, No. 14, 1997
Argazzi et al.
SCHEME 1
Experimental Section
Materials. All solvents and chemicals were of reagent grade
quality and used as received. RuCl3‚xH2O, 2,2′-bipyridine
(bpy), 2,2′-bipyridine-4,4′-carboxylic acid (dcb), 4,4′-dimethyl2,2′-bipyridine (dmb), NaI, and I2 were obtained from Aldrich.
10-Methylphenothiazine (MPTZ, Pfaltz and Bauer) and phenothiazine (PTZ , Aldrich) were recrystallized from toluene and
stored protected from light. Promethazine‚HCl (PMZ, Aldrich)
was used as received. Na4[(4,4′-(CO2-)2bpy)2Ru(Cl)2] was
available from previous studies.9b Ru(bpy)2Cl2 and Ru(bpy)2(dcb)(PF6)2 were synthesized according to literature preparations.15
TiO2 Preparations. Colloidal TiO2 solutions were prepared
by the technique of Micic et al.12 TiCl4 (2.2 mL) cooled to
-20 °C was added dropwise to cold water (0 °C) and then
dialyzed with nanopure water to a final pH of 2. The particles
were approximately 7 nm in diameter, and colloidal solutions
displayed negligible absorption or light scattering at wavelengths
longer than 400 nm. Titration with peroxide was used to
calculate the TiO2 concentrations as previously reported.12,13
Porous nanocrystalline anatase TiO2 electrodes were prepared
on fluorine-doped tin oxide conductive glass by deposition from
a concentrated colloidal TiO2 solution followed by a brief
sintering step as reported in the literature.14 Sintering at 450
°C in related materials causes necking between the particles
without significantly altering anatase particle size.7 The electrodes display little light scattering beyond 400 nm. The average
particle size is approximately 15 nm, and the final film thickness
is ∼10 µm. Dye attachment was achieved by soaking overnight
in 10-4 M sensitizer solutions.
Syntheses. [Ru(4-(COO-),4′-(COOH)-bpy)2(dmb)]. This
sensitizer was prepared by refluxing Na4[Ru(4,4′-(CO2-)2-bpy)2Cl2] (100 mg) with a stoichiometric amount of dmb in 60 mL
of 1/1 MeOH/H2O for 3 h in the dark under argon. The solution
was concentrated to ∼5 mL, loaded onto a silica column, and
eluted with NaCl saturated methanol. The first fraction was
collected, evaporated to dryness, and redissolved in water. The
neutral form was precipitated by addition of dilute HCl. Anal.
Calcd for RuC36Η26Ν6Ο8‚xΗ2Ο: C, 54.68; H, 3.57; N, 10.63.
Found: C, 56.7; H, 3.50; N, 9.60. FAB MS for Na2[Ru(4,4′-(CO2-)2-bpy)2(dmb)]: m/z ) 774 ([M + 1]).
[Ru(4-(COO-),4′-(COOH)-bpy)2(bpy-PTZ)]. This sensitizer
was prepared by the same route as [Ru(4-(COO-),4′-(COOH)bpy)2(dmb)] except that bpy-PTZ was substituted for dmb. Anal.
Calcd for RuC48H33N7O8S‚xH2O: C, 58.41; H, 3.57; N, 9.93.
Found: C, 56.7; H, 3.50; N, 9.60. FAB MS for Na2[Ru(4,4′-(CO2-)2-bpy)2(bpy-PTZ)]: m/z ) 1014 ([M + 1]).
Electrochemistry. All cyclic voltammetry was performed
in a one-compartment cell with an SCE reference electrode, a
Pt gauze counter electrode, and a Pt button working electrode.
A BAS CV-27 voltammogram was used for potential control
and wave form generation.
Photoelectrochemistry. Photoelectrochemical measurements
were performed in a two- and three-electrode arrangement in
dimethylformamide or propylene carbonate with 0.5 M NaI and
0.05 M I2. In a two-electrode arrangement, current and voltage
measurements were performed with a Keithley electrometer. In
a three-electrode arrangement, potential was controlled with a
BAS CV-27 voltammogram. Excitation sources were a 450
W Xe lamp coupled to a 0.22 m monochromator and the 488
or 514.5 nm line from an argon ion laser. Incident irradiance
was measured with a UDT-calibrated Si diode.
Open circuit photovoltages were measured in the absence of
iodide with 0.1 M tetrabutylammonium hexafluorophosphate
(TBAH) propylene carbonate electrolyte in both two- and threeelectrode arrangements. In the two-electrode arrangement, Voc
was measured with an electrometer vs a Pt reference electrode
or an Ag quasi reference electrode. In the three-electrode
arrangement, Voc was measured vs a silver wire quasi reference
electrode. In both configurations, some drift in the dark and
light voltages were observed. To minimize uncertainty, Voc
measurements were made immediately after closing the circuit
(two electrode) or potentiostatically adjusting the potential
difference to zero (three electrode).
Spectroscopy. UV-vis measurements were made on an HP
8451A diode array spectrometer. Corrected photoluminescence
Charge Separation across TiO2-Phenothiazine Interfaces
J. Phys. Chem. B, Vol. 101, No. 14, 1997 2593
(PL) spectra were obtained on a SPEX Model 112A fluorimeter.
Time-resolved PL measurements were performed with a previously described apparatus.16
Excited State Absorption Spectroscopy. Excited state
absorption spectroscopy was performed as previously described9b
with the following exceptions. The pump laser (frequencydoubled Nd:YAG, 532 nm, 5-7 ns fwhm) power was limited
to 5 mJ/pulse in all solution studies and to 10-12 mJ/pulse in
all thin film studies. In our study of TiO2 thin films and
colloidal solutions, significant light scattering was evident.
Kinetic data and excited state spectra of these samples were
obtained by using short and long pass filters in front of the
monochromator entrance slit that effectively blocked the scattered laser beam. In experiments with thin film electrodes, the
TiO2 surface was maintained at a 45° angle to the laser beam
and to the probe light. The laser power was monitored with a
Molectron PM-10V1 detector connected to a Molectron Power
Max 5200 laser power meter. The absorption transients were
plotted as ∆A ) log(Io/It) vs time, where Io was the monitoring
light intensity prior to the laser pulse and It was the observed
signal at time t after the laser pulse. Typically an average of
10 transients was used in kinetic analysis. Excited state
absorption spectra were plotted as ∆A vs wavelength at a
specified delay time after the laser pulse.
Fast Atom Bombardment Mass Spectroscopy. FAB MS
was obtained on a VG-Analytical Model 80 mass spectrometer.
FAB MS samples were suspended in a p-nitrobenzyl alcohol
matrix.
Results
The spectroscopic and redox properties of Ru(dcb)(bpy)22+
and Ru(dcb)2(dmb)2+ in pH 2 aqueous solution are typical of
MLCT excited states.17 The optical properties of Ru(dcb)2(bpyPTZ)2+ are consistent with previous reports of phenothiazineRu(II) donor-chromophore complexes.18 The photoluminescence (PL), absorption, and electrochemical properties of the
three sensitizers are tabulated in the Supporting Information.
Attachment of Ru(dcb)(bpy)22+ to TiO2 colloids was explored
spectroscopically at pH 2. The PL intensity from Ru(dcb)(bpy)22+* was quenched by the anaerobic addition of colloidal
TiO2. Under these same conditions, the Ru(dcb)(bpy)22+*
lifetime was constant, which is indicative of a static quenching
mechanism.19 The data are well described by the Stern-Volmer
model from which an adduct formation constant of Kad ) 20 (
2 M-1 was calculated. A slight broadening of the visible charge
transfer band of Ru(dcb)(bpy)22+ is observed upon addition of
TiO2.
Figure 1 shows the excited state absorption difference spectra
of a) the Ru(dcb)(bpy)22+ sensitizer, b) Ru(dcb)(bpy)22+/TiO2
assembly where greater than 80% of the PL intensity had been
quenched, and c) Ru(dcb)(bpy)22+/TiO2 with 10 mM PMZ. The
excited state absorption spectra of Ru(dcb)(bpy)22+ are typical
of Ru(II) polypyridyl compounds, and the kinetics are in good
agreement with those measured by time-resolved emission.
The excited state absorption difference spectra of Ru(dcb)(bpy)22+/TiO2 consists of contributions from the unquenched
MLCT excited states and from RuIII(dcb)(bpy)23+. Under no
conditions were absorption transients observed which might be
assigned to electrons in TiO2. Time-resolved absorbance
changes measured at the isosbestic point (397 nm) between the
MLCT excited state and the ground state allow formation and
reduction of the oxidized sensitizer to be cleanly observed,20
Figure 2, part a. The growth of this absorption feature could
not be time resolved, which indicates that electron injection
occurs within the laser pulse, k2 > 5 × 107 s-1. The transient
Figure 1. Excited state absorption spectra of the following assemblies
at pH 2 recorded after excitation with a 532 nm laser pulse: (a) Ru(dcb)(bpy)22+ at (circles) 20 ns, (squares) 200 ns, and (diamonds) 500
ns. (b) Same as (a) with the addition of 80 mM TiO2 colloid (circles)
50 ns, (squares) 500 ns, and (diamonds) 5 µs. (c) Same as (b) except
the addition of 20 mM PMZ (circles) 50 ns, (squares) 2 µs, and
(diamonds), 45 µs.
displays complex kinetics, but recovers cleanly to base line on
a microsecond time scale. The kinetics were well described
by the Kohlrausch-Williams-Watts (KWW) function,
eq 1.21 A mean rate constant 〈k〉 was calculated with eq 2 where
β
( τt ) 0 < β < 1
τ 1
〈k〉 ) [( )Γ( )]
β β
∆A(t) ) R exp -
(1)
-1
(2)
Γ represents the gamma function. While the KWW function
analytically describes the observed kinetics here and elsewhere,20
we note that the inverse Laplace transform results in physically
unrealistic distributions of first-order rate constants with significant amplitude over 6 decades of rate space.11b The mean
rates calculated do provide a basis for internal quantitative
comparisons of different assemblies.
The excited state absorption spectra in Figure 1, part c shows
the appearance of a new feature with a maximum at 520 nm,
assigned to PMZ+. The assignment is based on previous reports
2594 J. Phys. Chem. B, Vol. 101, No. 14, 1997
Argazzi et al.
TABLE 1: Intermolecular Electron Transfer Kinetics with
Phenothiazinesa
assembly
Ru(dcb)(bpy)22+ b
Ru(dcb)(bpy)22+/TiO2 (colloid)
Ru(dcb)(bpy)22+/TiO2 (film)
Ru(dcb)(bpy)22+/TiO2 (film)
Ru(dcb)2(dmb)2+/TiO2 (film)
donor
k4 × 10-8
(M-1 s-1)
k5 × 10-8
(M-1 s-1)
PMZ 14 ( 3
80 ( 5
PMZ
2.9 ( 0.5 0.134 ( 0.005
PMZ
3.2 ( 0.2 2.0 ( 0.3c
MPTZ 4.2 ( 0.7 0.83 ( 0.1c
MPTZ 2.6 ( 0.5 87 ( 6c
a
Electron transfer kinetics for the indicated assemblies where k4 and
k5 refer to the processes indicated in Scheme 1 unless otherwise noted.
Abbreviations for the sensitizer and donor are given in the text. The
kinetics measured for assemblies listed as (colloid) were measured in
pH 2 colloidal solution and those listed as (film) were measured in
propylene carbonate. The error given represents one standard deviation
from multiple samples. b The k4 and k5 designations do not apply to
this assembly since TiO2 is not present. Instead, they refer to direct
quenching of the MLCT excited state (under k4) and subsequent
recombination to ground state products (under k5). c A pathlength of
ten µm was employed to calculate this rate.
Figure 2. Single-wavelength absorbance changes for the assemblies
described in Figure 1 after excitation with a 532 nm laser pulse. (a)
Kinetics observed at 397 nm for Ru(dcb)(bpy)22+ in colloidal TiO2
solutions at pH 2. The residuals are for fits to the KWW model. (b)
Kinetics observed at 510 nm recorded to 2 µs for Ru(dcb)(bpy)22+ in
colloidal TiO2 solutions at pH 2 with 20 mM PMZ. The inset is a plot
of the observed first-order rate constant as a function of PMZ
concentration from which the second-order rate constant is calculated,
2.9 × 108 M-1 s-1. (c) Kinetics observed at 510 nm recorded for the
same assembly as (b) except recorded to 200 ms. The inset shows a
second-order equal concentration kinetics analysis that yields a rate
constant of 1.34 × 107 M-1 s-1.
and the appearance of the same feature following direct (λexc
) 355 nm) excitation of PMZ in pH 2 water. A complication
that is not shown in Scheme 1 is that the phenothiazine
derivatives can reduce both the MLCT and the oxidized state
of the sensitizers. Quenching of the MLCT excited states may
be particularly significant in colloidal TiO2 solutions where the
yield of electron injection is only ∼80%. However, reductive
quenching of Ru(dcb)(bpy)22+* by PMZ at pH 2 is consistently
an order of magnitude faster than that observed for the sensitizer
anchored to TiO2. Further, reductive quenching of the MLCT
excited state yields Ru(dcb)(bpy)2+ which is not observed
spectroscopically 200 ns after laser excitation at high PMZ
concentrations. In all cases, the appearance of the 520 nm
absorption feature follows pseudo-first-order kinetics from which
the second-order rate k4 can be abstracted, Figure 2, part b. The
decay of PMZ+ occurs on a millisecond time scale and follows
second-order equal concentration kinetics, k5, shown in Figure
2, part c. The kinetics are summarized in Table 1.
The electron transfer studies described above were repeated
for Ru(dcb)(bpy)22+ anchored to porous nanocrystalline TiO2
films. The sensitizer was found to be unstable under these
conditions so the electron transfer measurements were performed
in 0.1 M LiClO4 propylene carbonate electrolyte with MPTZ
as the electron donor. Excited state absorption spectra were
qualitatively the same as those observed in colloidal solutions,
and the kinetics were well described by the same kinetic models
described for the colloidal solutions, Figure 3. An experimental
difficulty came in quantifying k5 in the colloidal films. The
absorbance change follows a second-order equal concentration
kinetic model as shown by the residuals in Figure 3, part c.
However, the appropriate path length is unknown, which results
in significant uncertainty in the calculated rate. In an effort to
limit the path length, the kinetics were measured in a “sandwich”
cell arrangement with a few drops of MPTZ in propylene
carbonate sandwiched between the TiO2 films and a piece of
glass. This restricts the path length to approximately that of
the TiO2 film, ∼10 µm, but may introduce some bias into the
calculated rates given in Table 1. While the poor stability of
the assembly in pH 2 water precluded signal averaging, singleshot data demonstrate that the kinetics were not dramatically
different than those measured in propylene carbonate. The
kinetics were not altered significantly if PMZ or PTZ was used
in place of MPTZ or when more LiClO4 was added.
[Ru(dcb)2(bpy-PTZ)]2+ and Ru(dcb)2(dmb)2+ anchored to
TiO2 electrodes display excited state absorption difference
spectra shown in Figure 4. The difference spectra of Ru(dcb)2(dmb)2+/TiO2 strongly resembles that of Ru(dcb)(bpy)22+/TiO2.
The formation of the oxidized sensitizer cannot be resolved at
the isosbestic point, 410 nm. The recovery is well described
by the KWW model, 〈k3〉 ) 3.9 × 106 s-1. The excited state
absorption spectra of [Ru(dcb)2(bpy-PTZ)]2+/TiO2 at 20 ns
displays a positive absorption band assigned to the oxidized
phenothiazine radical, Figure 4, part b. There is evidence for
a bleach of the charge transfer bands that cannot be kinetically
resolved with our instrumentation. The loss of the 510 nm
feature follows first-order kinetics, k5 ) 3.6 × 103 s-1.
The photoelectrochemical properties of Ru(dcb)(bpy)22+/TiO2
were explored in propylene carbonate 0.1 M TBAH electrolyte
with MPTZ as the donor. In the best case scenarios, the incident
photon-to-current-efficiencies (IPCE) were 6 × 10-4 at 460 nm.
Addition of 0.1 M LiClO4 or MPTZ+ had little effect on the
IPCE.
Charge Separation across TiO2-Phenothiazine Interfaces
J. Phys. Chem. B, Vol. 101, No. 14, 1997 2595
Figure 3. Absorbance kinetics analogous to those in Figure 2, except
that a colloidal TiO2 film was used with MPTZ donors in 0.1 M LiClO4
propylene carbonate electrolyte. (a) Kinetics observed at 397 nm for
Ru(dcb)(bpy)22+/TiO2. (b) Kinetics observed at 510 nm recorded to 1
µs for Ru(dcb)(bpy)22+/TiO2 with 20 mM MPTZ. The inset is a plot of
the observed first-order rate constant as a function of MPTZ concentration from which the second-order rate constant is calculated, 4.2 ×
108 M-1 s-1. (c) Kinetics observed at 510 nm recorded for the same
assembly as (b) except recorded to 200 ms. The inset shows residuals
to a second-order equal concentration kinetics model. This analysis
yields a rate constant of 8.3 × 107 M-1 s-1.
The photoelectrochemical properties of the three sensitizers
anchored to TiO2 with iodide as an electron donor were explored
in a two-electrode geometry. Short circuit photocurrents isc and
open circuit photovoltages Voc measured at different excitation
irradiances are well described by eq 3 with an ideality factor n
Voc )
( )()
isc
nkT
ln
e
io
(3)
of approximately 2. Equation 3 implicitly assumes that the
saturation current io is much smaller than that of isc which was
true at all the light irradiances employed. Plots of Voc vs log
irradiance are linear over 4 decades of irradiance. The slopes
of typical plots were 100 ( 10 mV/decade in the presence of
iodide and 60 ( 20 mV/decade in the absence of iodide for all
three sensitizers. These data are shown in the Supporting
Information.
Figure 4. (a) Excited state absorption difference spectra of Ru(dcb)2(dmb) anchored to a transparent TiO2 film in propylene carbonate
electrolyte. Spectra are shown 50 ns (circles) and 500 ns (squares) after
excitation with a 10 mJ, 5 ns pulse of 532 nm light. The data represent
the average of 10 laser pulses. Kinetics of the absorbance at 410 nm
are shown in the inset. Analysis based on the KWW model yields an
average rate, 〈k3〉 ) 3.91 × 106 s-1. (b) Excited state absorption
difference spectra of Ru(dcb)2(bpy-PTZ) under the same conditions as
those in (a). Spectra are shown 500 ns (circles) and 5 µs (squares).
The inset depicts the kinetics measured at 514 nm, assigned to the
recombination of theTiO2(e-)|-RuII-PTZ+ charge-separated pair
recorded with ∼2 µm resolution. A first-order kinetics analysis yields
k5 ) 3.6 × 103 s-1.
Discussion
Surface-anchored Ru(dcb)(bpy)22+* injects an electron into
TiO2 by a static mechanism with a rate constant that cannot be
resolved by our instrumentation, k2 > 5 × 107 s-1. Rapid
electron injection is consistent with recent microwave22 and
photoluminescence23 studies that reveal a distribution of electron
injection rates with a mean value of ∼108 s-1. Therefore, with
reference to Scheme 1, only those processes to the right of the
broad arrow could be kinetically resolved in this study. The
recombination of the electron in the solid with the oxidized form
of the sensitizer occurs with average rates on a microsecond
time scale, k3 ≈ 105-106 s-1, consistent with many other
reports.24 The complex kinetics presumably reflect interfacial
heterogeneity.
Phenothiazine derivatives in fluid solution efficiently intercept
the interfacial charge-separated state and reduce the oxidized
sensitizer before recombination, k4 in Scheme 1. Significantly,
the efficiency and rate of this process are the same for colloidal
solutions and thin films. This indicates that the PTZ derivatives
have access to all the Ru(III) sites in the nanostructured film.
In colloidal TiO2 solutions, the charge-separated pairs [TiO2(e-),
2596 J. Phys. Chem. B, Vol. 101, No. 14, 1997
SCHEME 2
Argazzi et al.
based on crystalline Si materials in great detail.28 They find
that Voc is not a thermodynamic quantity, but rather a kinetic
variable of a photostationary state. For an n-type semiconductor,
the open circuit voltage is the potential at which the majority
carrier current density due to electron injection from the
conduction band (Iinj) exactly offsets the photogenerated interfacial hole current density from the valence band. Equations 3
and 4, and modified forms often referred to as diode equations,
Voc )
PMZ+] recombine remarkably slowly, 1.34 ( 0.05 × 107 M-1
s-1. Recombination is 3 orders of magnitude faster in the
absence of TiO2, i.e., [Ru(dcb)(bpy)2+, PMZ+], under the same
conditions. This dramatic increase in charge-separated state
lifetime is achieved by simply anchoring the sensitizer to
colloidal TiO2 particles.
Unfortunately, efficient recombination of electrons in TiO2
colloidal films with oxidized phenothiazine compounds results
in low incident photon-to-current-efficiencies in regenerative
solar cells. The high local concentrations present in the colloidal
films significantly shorten the lifetimes of oxidized phenothiazines, and they do not escape from the complex threedimensional TiO2 network. Attempts to increase chargeseparated state lifetimes by changing the solvent, ionic strength,
and/or the phenothiazine donor have been unsuccessful. New
strategies which block the back reaction to ground state products
are required before efficient solar cells based on phenothiazine
donors can be realized. It is interesting to note that previous
attempts to employ other outer sphere one-electron donors in
regenerative solar cells of this type have also led to low IPCE.25
One strategy to increase charge separation lifetimes is to
covalently bond the donor to the Ru(II) sensitizer. Less than
20 ns after excitation of surface-anchored [Ru(dcb)2(bpyPTZ)]2+, an electron is injected into TiO2 and the PTZ group
reduces the metal center. It is unclear which happens first,
Scheme 2. These excited state electron transfer reactions
produce interfacial charge-separated pairs which are remarkably
long-lived, ket ) 3.6 × 103 s-1.
When employed as a photoanode in a regenerative solar cell
with iodide as an electron donor, [Ru(dcb)2(bpy-PTZ)]2+/TiO2
efficiently converts photons into electrons. For a large number
of samples, the IPCE is 45 ( 10%, which is within experimental
error the same as that observed for Ru(dcb)2(dmb)2+/TiO2 and
Ru(bpy)2(dcb)2+/TiO2 photoanodes. A key difference in the
photoelectrochemical properties, however, is an ∼100 mV larger
open circuit photovoltage, Voc, for [Ru(dcb)2(bpy-PTZ)]2+/TiO2
when compared to that for Ru(dcb)2(dmb)2+/TiO2, which serves
as a model. The factors which govern open circuit photovoltages, Voc, in these devices are not well understood. Voc defines
the maximum Gibbs free energy that can be obtained from a
photoelectrochemical cell under constant light irradiance conditions.26 The maximum open circuit photovoltage attainable is
the energetic difference between the Fermi level of the solid
under illumination and the Nernst potential of the redox couple
in the electrolyte. However, this maximum has not been
realized, and there is growing evidence that Voc is kinetically
limited by electron tunneling through the solid to the oxidized
form of the dye and/or the electron donor.27
Lewis and co-workers have examined the factors that control
open circuit voltages in regenerative photoelectrochemical cells
()( )
kT
e
Iinj
ln
∑i ki[A]i
(4)
n
are applicable where n is the concentration of electrons in the
semiconductor and ki is the electron transfer rate to acceptor
Ai.
The diode equation is based on a Schottky junction model
of the semiconductor interface. There is no a priori reason to
expect that these nanostructured interfaces should behave like
ideal diodes or that this equation should be valid. In fact, there
are very good reasons to believe that they should not.6,7
Nevertheless, these interfaces clearly rectify charge, and diode
equations have been successfully applied to related nanostructured materials.6,29 In regenerative solar cells with iodide
donors, the diode equation accurately describes the currentvoltage behavior with an ideality factor of ∼2 in agreement
with previous reports.6 It is not possible to construct meaningful
current-voltage curves in the absence of iodide; however, the
slope of the Voc vs log(irradiance) indicates that the interface
behaves like an ideal diode. Further, we have shown that eq 4
accurately predicts open circuit voltage differences in the
absence of iodide.11b The results presented here support this
conclusion over 4 decades of irradiance. If one assumes that
Iinj is the same for [Ru(dcb)2(bpy-PTZ)]2+ and Ru(dcb)2(dmb)2+,
then the measured interfacial kinetics and eq 4 predict a 200
mV larger Voc for [Ru(dcb)2(bpy-PTZ)]2+/TiO2 compared to that
for Ru(dcb)2(dmb)2+/TiO2, which agrees well with the measured
values, Voc ) 180 ( 30 mV.
The apparent applicability of the diode equation to nanometersized semiconductor particles that cannot support large electric
fields can be rationalized if an exponential distribution of states
exists in the material. An exponential relation between dark
current and potential would then follow, and the diode equation
would be valid. The density of states in the TiO2 colloidal films
remains unknown; however, an exponential distribution was
recently proposed to model the voltammetry of TiO2 observed
at negative applied potentials.30 While the data here support
this model, we emphasize that the density of states is unknown
and a tail of states from many distributions might well
approximate an exponential distribution. Further studies are
clearly needed before this interesting optoelectronic phenomena
can be fully rationalized.
Conclusion
In conclusion, phenothiazine donors increase the lifetime of
interfacial charge-separated pairs based on colloidal TiO2
solutions with Ru(II) sensitizers. Efficient charge recombination
in corresponding colloidal TiO2 films results in very low IPCE
for regenerative solar cells. A novel sensitizer has been
designed to vectorially translate the oxidizing equivalent (the
hole) away from the nanostructured semiconducting interface.
The decreased electronic coupling between the surface and the
hole results in a remarkably long-lived charge-separated pair
which directly leads to an increased open circuit photovoltage.
Further, the molecular level recombination kinetics applied to
Charge Separation across TiO2-Phenothiazine Interfaces
a solid state model quantitatively predict the increased efficiency.
The general strategy of vectorial translation of photogenerated
holes away from interfaces is successful and may be applied to
other assemblies to prevent charge recombination and increase
solar conversion efficiencies.
Acknowledgment. We thank the National Renewable Energy Laboratory (NREL XAD-3-12113-04), the National Science Foundation (CHE-9322559, CHE-9402935), and MURST
for support of this research. T.A.H. acknowledges support from
a Sonneborn fellowship.
Supporting Information Available: Summary of the electrochemical and spectroscopic properties of the sensitizers,
Stern-Volmer analysis of the PL quenching by colloidal TiO2,
and photoelectrochemical data for these materials in the presence
and absence of iodide (3 pages). Ordering information is given
on any current masthead page.
References and Notes
(1) (a) Gerischer, H. Pure Appl. Chem. 1980, 52, 2649. For more recent
reviews, see: (b) Hagfeldt, A.; Grätzel, M. Chem. ReV. 1995, 95, 49. (c)
Kamat, P. V. Prog. React. Kinet. 1994, 19, 277. (d) Henglein, A. Top. Curr.
Chem. 1988, 143, 115.
(2) (a) Gerischer, H. J. Phys. Chem. 1991, 95, 1356. (b) Fox, M. A.;
Chanon, M. Photoinduced Electron Transfer; Elsevier: Amsterdam, 1988.
(c) Gratzel, M. Heterogeneous Photochemical Electron Transfer; CRC:
Boca Raton, FL, 1989.
(3) Meyer, G. J.; Searson, P. C. Interface 1993, 2, 23.
(4) Hodes, G.; Howell, I. D.; Peter, L. M. J. Electrochem. Soc. 1992,
139, 3136.
(5) Rothenberger, G.; Grätzel, M.; Fitzmaurice, D. J. Phys. Chem. 1992,
96, 5983.
(6) Sodergren, S.; Hagfeldt, A.; Olsson, J.; Lindquist, S.-E. J. Phys.
Chem. 1994, 98, 5552.
(7) (a) Cao, F.; Oskam, G.; Searson, P. C.; Stipkala, J. M.; Heimer, T.
A.; Farzad, F.; Meyer, G. J. J. Phys. Chem. 1995, 99, 11974. (b) Cao, F.;
Oskam, G.; Searson, P. C.; Meyer, G. J. J. Phys. Chem. 1996, 100, 17021.
(8) (a) Desilvestro, J.; Grätzel, M.; Kavan, L.; Moser, J.; Augustynski,
J. J. Am. Chem. Soc. 1985, 107, 2988. (b) O’Regan, B.; Grätzel, M. Nature
1991, 353, 737. (c) Nazeerudin, M. K.; Kay, A.; Rodicio, I.; Humphry, B.
J. Phys. Chem. B, Vol. 101, No. 14, 1997 2597
R.; Mueller, E.; Liska, P.; Vlachopoulos, N.; Grätzel, M. J. Am. Chem.
Soc. 1993, 115, 6382.
(9) (a) Heimer, T. A.; Bignozzi, C. A.; Meyer, G. J. J. Phys. Chem.
1993, 97, 11987. (b) Argazzi, R.; Bignozzi, C. A.; Heimer, T. A.; Castellano,
F. N.; Meyer, G. J. Inorg. Chem. 1994, 33, 5741.
(10) Wardman, P. J. Phys. Chem. Ref. Data 1989, 18, 1704.
(11) (a) Argazzi, R.; Bignozzi, C. A.; Heimer, T. A.; Castellano, F. N.;
Meyer, G. J. J. Am. Chem. Soc. 1995, 117, 11815. (b) Heimer, T. A.;
Castellano, F. N.; Meyer, G. J. Abstracts of Papers, 211th National Meeting
of the American Chemical Society, New Orleans, LA, Spring, 1996;
American Chemical Society: Washington, DC, 1996; PHYS 138.
(12) Micic, O. I.; Zhang, Y.; Cromack, K. R.; Trifunac, A. D.; Thurnauer,
M. C. J. Phys. Chem. 1993, 97, 7277.
(13) Ellis, J. D.; Sykes, A. G. J. Chem. Soc., Dalton Trans. 1973, 537.
(14) Heimer, T. A.; D’Arcangelis, S. T.; Farzad, F.; Stipkala, J. M.;
Meyer, G. J. Inorg. Chem. 1996, 35, 5319.
(15) Shimidzu, T.; Iyoda, T.; Izaki, K. J. Phys. Chem. 1985, 89, 642.
(16) Castellano, P. C.; Heimer, T. A.; Thandasetti, M.; Meyer, G. J.
Chem. Mater. 1994, 6, 1041.
(17) For recent reviews of MLCT excited states, see: (a) Meyer, T. J.
Acc. Chem. Res. 1989, 22, 364. (b) Balzani, V.; Scandola, F. Supramolecular
Photochemistry; Ellis Harwood: Chichester, U.K., 1990.
(18) Larson, S. L.; Elliott, C. M.; Kelley, D. F. Inorg. Chem. 1996, 35,
2070.
(19) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Plenum
Press: New York, 1983.
(20) Ford, W. E.; Rodgers, M. A. J. J. Phys. Chem. 1994, 98, 3822.
(21) (a) Kohlrausch, R. Ann. 1847, 5, 430. (b) Williams, G.; Watts, D.
C. Trans. Faraday Soc. 1971, 66, 80.
(22) Fessenden, R.; Kamat, P. V. J. Phys. Chem. 1995, 99, 12902.
(23) Heimer, T. A.; Meyer, G. J. J. Lumin. 1996, 70, 468.
(24) (a) Desilvestro, J.; Grätzel, M.; Kavan, L.; Moser, J.; Augustynski,
J. J. Am. Chem. Soc. 1985, 107, 2988. (b) O’Regan, B.; Moser, J.; Anderson,
M.; Grätzel, M. J. Phys. Chem. 1990, 94, 8720.
(25) Bonhote, P.; Grätzel, M.; Jirousek, M.; Liska, P.; Pappas, N.;
Vlachopoulos, N.; von Plata, C.; Walder, L. Int. Conf. Photochem. ConVers.
& Storage Sol. Energy, 10th 1994, Abstract C2.
(26) Fahrenbruch, A. L.; Bube, R. H. Fundamentals of Solar Cells
PhotoVoltaic Solar Energy ConVersion, Academic Press: New York, 1983.
(27) Lindstrom, H.; Rensmo, H.; Sodergren, S.; Solbrand, A.; Lindquist,
S. E. J. Phys. Chem. 1996, 100, 3084.
(28) Tan, M. X.; Laibinis, P. E.; Nguyen, S. T.; Kesselman, J. M.;
Stanton, C. E.; Lewis, N. S. Prog. Inorg. Chem. 1994, 41, 21 and references
therein.
(29) Hotchandani, S.; Kamat, P. V. J. Electrochem. Soc. 1992, 139, 1630.
(30) Kay, A.; Humphry-Baker, R.; Grätzel, M. J. Phys. Chem. 1994,
98, 952.